Vegetation coverage estimation method and device based on remote sensing semantic segmentation

Abstract

The invention provides a vegetation coverage estimation method and device based on remote sensing semantic segmentation. The method comprises the following steps: acquiring a remote sensing image of a to-be-detected region; inputting the remote sensing image into a preset deep learning model to obtain a vegetation classification result; and carrying out statistical analysis on the classification result, and calculating the vegetation coverage. The remote sensing image is input into the deep learning model, and the model combines depth separable convolution and residual connection through multi-path feature fusion and SiLU activation function introduction, so that the model parameter quantity and the calculation cost are reduced, the feature extraction capability is enhanced, the precision of the segmentation result and the detail retention capability are improved, and the segmentation efficiency is improved. The method shows good non-linear expression capability when processing a multi-scale remote sensing image, and effectively improves the estimation precision of complex ground features and vegetation coverage. The segmentation precision and the calculation efficiency of the remote sensing image are greatly improved, a solid technical support is provided for efficient and accurate vegetation coverage estimation, and the method has a wide application prospect.

Description

Technical Field

[0001] The present application relates to the technical field of vegetation balanced coverage prediction, and in particular to a vegetation coverage estimation method and device based on remote sensing semantic segmentation. Background Art

[0002] Fractional vegetation cover (FVC) is an important indicator for assessing ecosystem health. It has a significant impact on environmental protection, climate change mitigation, biodiversity conservation and sustainable development. It is also a key indicator for assessing plant growth and ecological environmental changes. This indicator has been applied in many fields, including analyzing the physiological state of plants, monitoring crop growth and estimating food output, as well as vegetation degradation and land desertification assessment, disaster monitoring, soil conservation capacity and other research. Traditional sample measurement methods are widely used, but they are highly subjective and inefficient. With the advancement of science and technology, remote sensing technology has gradually become the main means of measuring vegetation cover.

[0003] Existing remote sensing vegetation coverage estimation methods mainly include vegetation index-based methods, such as the Normalized Difference Vegetation Index (NDVI) and the Enhanced Vegetation Index (EVI). These methods use the information of different spectral bands of remote sensing images to generate vegetation indices through simple mathematical operations to reflect the growth status and coverage of vegetation. However, they are affected by spectral mixing and environmental factors and have limited accuracy. The calculation of vegetation index is based on the reflectivity difference between visible light and near-infrared light bands. Because of its simple calculation and easy large-scale application, it is widely used in ecological monitoring and agricultural management. These methods have spectral mixing phenomena, which makes it difficult to distinguish different objects in the same pixel, thereby reducing accuracy.

[0004] Another commonly used method is the classification method based on threshold segmentation. This method classifies pixels into vegetation and non-vegetation by setting specific spectral thresholds, such as pixel dichotomy. This method can simplify the processing process to a certain extent, but it is highly dependent on image quality and user experience and is easily affected by human factors, resulting in insufficient reliability of the results. The model is not adaptable enough to different vegetation types and ecological environments, which may lead to poor performance in specific areas and ultimately affect the reliability of the overall estimate. Summary of the invention

[0005] The present application provides a method and device for vegetation coverage estimation based on remote sensing semantic segmentation from multiple aspects, constructs a remote sensing semantic segmentation network framework based on an encoder-decoder architecture, realizes lightweight and efficient vegetation coverage estimation, and improves the efficiency and accuracy of vegetation coverage estimation.

[0006] To achieve the above technical effects, one aspect of the present application provides a vegetation coverage estimation method based on remote sensing semantic segmentation, comprising:

[0007] Acquire remote sensing images of the area to be measured;

[0008] Inputting the remote sensing image into a preset deep learning model to obtain a vegetation classification result;

[0009] The classification results were statistically analyzed and the vegetation coverage was calculated.

[0010] According to a preferred embodiment of the present invention, inputting the remote sensing image into a preset deep learning model to obtain a vegetation classification result further comprises:

[0011] Build a convolutional neural network model based on the encoder-decoder architecture;

[0012] Inputting the remote sensing image into the convolutional neural network model for semantic segmentation;

[0013] According to the semantic segmentation result, the vegetation type in the remote sensing image is output.

[0014] According to a preferred embodiment of the present invention, inputting the remote sensing image into the convolutional neural network model for semantic segmentation further comprises:

[0015] Input the remote sensing image into the Stem module for preliminary feature extraction to generate a shallow feature map;

[0016] The shallow feature map is subjected to convolution calculation through multiple Block modules to extract image features;

[0017] The processed feature map is restored to its original resolution and the semantic segmentation result is generated.

[0018] According to a preferred embodiment of the present invention, the step of performing convolution on the shallow feature map through multiple Block modules to calculate image features further includes:

[0019] Normalize the input shallow feature map;

[0020] The activation function SiLU is used to perform convolution calculation on the normalized features to extract image features.

[0021] According to a preferred embodiment of the present invention, the step of performing convolution calculation on the normalized features using the activation function SiLU to extract image features further comprises:

[0022] In the upper path, the normalized features are linearly transformed and convolved, and the SiLU activation function is used to obtain the upper path features;

[0023] In the lower path, the normalized features are linearly transformed and the depth-wise separable convolution DWConv is calculated, and the SiLU activation function is used to obtain the lower path features;

[0024] The upper layer path features and the lower layer path features are respectively residually connected and added to the original input features to obtain complete upper layer path features and lower layer path features.

[0025] The complete upper-layer path features and lower-layer path features are added and fused element by element to obtain the image features.

[0026] According to a preferred embodiment of the present invention, performing statistical analysis on the classification results to calculate the vegetation coverage further includes:

[0027] The proportion of each type of vegetation in the remote sensing image is calculated, and the overall vegetation coverage in the remote sensing image is calculated.

[0028] According to a preferred embodiment of the present invention, the method further comprises:

[0029] The training samples of the deep learning model are updated regularly, and the model parameters are updated using the updated training samples.

[0030] Another aspect of the present application provides a vegetation coverage estimation device based on remote sensing semantic segmentation, comprising:

[0031] Image acquisition module: used to obtain remote sensing images of the area to be measured;

[0032] Image classification module: used to input the remote sensing image into a preset deep learning model to obtain vegetation classification results;

[0033] Statistical analysis module: used to perform statistical analysis on classification results and calculate vegetation coverage.

[0034] Another aspect of the present application provides a vegetation coverage estimation device based on remote sensing semantic segmentation, the device comprising:

[0035] at least one processor; and

[0036] a memory communicatively connected to the at least one processor; wherein,

[0037] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method described above.

[0038] In another aspect of the present application, a computer-readable storage medium is provided, on which computer program instructions are stored. The computer program instructions can be executed by a processor to implement the above method.

[0039] In the solution provided in the embodiment of the present application, by adopting a lightweight design, the network greatly reduces the computational complexity while ensuring the segmentation accuracy, making it suitable for devices with limited resources and improving the flexibility of practical applications. Secondly, the Block module combines deep separable convolution and residual connection, which not only reduces the number of model parameters and computational cost, but also enhances the ability of feature extraction, and improves the accuracy and detail retention of segmentation results. Through the introduction of multi-path feature fusion and SiLU activation function, the network shows good nonlinear expression ability when processing multi-scale remote sensing images, and effectively improves the estimation accuracy of complex objects and vegetation coverage. In addition, the application of layer normalization significantly enhances the stability of the model, ensuring that it can converge efficiently under different data distributions. In general, the present invention greatly improves the segmentation accuracy and computational efficiency of remote sensing images, provides solid technical support for efficient and accurate vegetation coverage estimation, and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0041] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0042] Figure 1 A schematic diagram of a process flow of a vegetation coverage estimation method based on remote sensing semantic segmentation provided in an embodiment of the present application;

[0043] Figure 2 A schematic diagram of a deep learning model structure provided in one embodiment of the present application;

[0044] Figure 3 A schematic diagram of a flow chart of extracting image features by a Block module of a deep learning model provided in one embodiment of the present application;

[0045] Figure 4 A comparison chart of the vegetation coverage estimation method based on remote sensing semantic segmentation provided in an embodiment of the present application and the visualization results of the prior art;

[0046] Figure 5 A comparison chart of the vegetation coverage estimation method based on remote sensing semantic segmentation and the visualization result of the pixel dichotomy method provided in an embodiment of the present application;

[0047] Figure 6A schematic diagram of a vegetation coverage estimation device based on remote sensing semantic segmentation provided in one embodiment of the present application;

[0048] Figure 7 The present invention is a schematic diagram of the structure of a device suitable for implementing the solution in the embodiment of the present application.

[0049] The same or similar reference numerals in the drawings represent the same or similar components. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0051] In a typical configuration of the present application, the terminal and the equipment of the service network each include one or more processors (CPU), input / output interface, network interface and memory.

[0052] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0053] Computer readable media include permanent and non-permanent, removable and non-removable media, and can be implemented by any method or technology to store information. Information can be computer program instructions, data structures, modules of programs or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0054] In actual scenarios, the execution subject of the method can be a user device, or a device formed by integrating a user device and a network device through a network, or an application running on the above device. The user device includes but is not limited to various terminal devices such as computers, mobile phones, tablet computers, smart watches, and bracelets. The network device includes but is not limited to network hosts, single network servers, multiple network server sets, or cloud computing-based computer sets, which can be used to implement some processing functions when setting an alarm. Here, the cloud is composed of a large number of hosts or network servers based on cloud computing, where cloud computing is a type of distributed computing, a virtual computer composed of a group of loosely coupled computer sets.

[0055] In recent years, the application of machine learning and deep learning techniques has made significant progress in remote sensing vegetation cover estimation. By training models to automatically learn features, these methods can handle complex nonlinear relationships and improve the ability to recognize diverse objects. In particular, convolutional neural networks (CNNs) and semantic segmentation techniques make it possible to accurately classify at the pixel level, further improving the accuracy and efficiency of remote sensing image processing, and are suitable for complex object scenes and multi-source data fusion.

[0056] Figure 1 A flowchart of a vegetation coverage estimation method based on remote sensing semantic segmentation is provided in an embodiment of the present application, such as Figure 1 As shown, the method includes:

[0057] S101, obtaining a remote sensing image of the area to be measured.

[0058] Specifically, remote sensing information of the area to be measured is obtained from the Internet or remote sensing images collected by satellites. The remote sensing information includes at least forest coverage area information, forest density information, land coverage area information and specific ecological indicators. Forest coverage area refers to the ground area covered by forests in a specified area. Through time series data, the changing trend of forest coverage area can be observed to understand the expansion or contraction of forests; forest density is usually used to describe the density of trees in the forest, which can be the number of trees per unit area and the canopy coverage, which can reveal the health status, growth rate and possible interference of the forest; land coverage area represents the total area of ​​various land cover types (such as grassland, water area, bare land, construction land) in a specified area, which can help understand the changes in land use, such as the impact of urbanization on land cover; specific ecological indicators can be biodiversity index, vegetation index (such as NDVI), soil erosion degree, which can reflect the key information of a certain aspect of the ecosystem.

[0059] S102: Input the remote sensing image into a preset deep learning model to obtain a vegetation classification result.

[0060] Specifically, we first build a convolutional neural network model based on the encoder-decoder architecture. Figure 2 A schematic diagram of a deep learning model structure provided in an embodiment of the present application is shown in FIG. Figure 2 As shown in the figure, the model includes a Stem module, a Block module, and a decoding module (Decoder). The Stem module is used as the entrance of the network for preliminary feature extraction and generating shallow feature maps. It uses a lightweight convolutional layer combination to effectively extract shallow features of the input image through a series of convolution, pooling, and activation operations, emphasizing the balance between efficiency and feature extraction capabilities, ensuring that the subsequent Block module can receive high-quality basic feature maps. In addition, by using a smaller filter size and step size, the Stem module can maintain a high spatial resolution, which is crucial for subsequent high-level feature learning.

[0061] As the core component of the network, the Block module adopts a multi-branch structure for performing convolution calculations to extract image features. Each Block module combines depthwise separable convolution to reduce the number of parameters and computational costs, while introducing residual connections to retain shallow detail features, which helps to improve segmentation accuracy. Through the downsampling layer, the network extracts multi-scale features at different depths to adapt to objects with multi-scale characteristics in remote sensing images. In order to ensure high efficiency and rational use of computing resources, the present invention significantly reduces the amount of calculation of the convolution layer through depthwise separable convolution, while reducing the number of parameters while maintaining model accuracy, and is suitable for running on devices with limited resources. In addition, the combination of SiLU (Sigmoid-weighted Linear Unit) activation function and layer normalization enables the network to have good nonlinear expression ability and stability in the process of remote sensing image processing. Layer normalization effectively alleviates the problem of gradient explosion or disappearance, ensuring that the network can converge stably in complex data distribution scenarios.

[0062] The decoding module is used to gradually restore the upsampling operation of the feature map after multi-layer processing to the original resolution and generate the final semantic segmentation result. At the same time, it integrates the high-level semantic information from the encoding stage with the low-level detail features to ensure that the output segmentation result is both accurate and rich in details.

[0063] After building a convolutional neural network model based on the encoder-decoder architecture, the original data required for a county was obtained from the remote sensing data center, and image blocks of 512x512 pixels were obtained by segmentation. In order to train the deep learning model, the data needs to be accurately labeled, and each image block is labeled with: level I, level II, level III, level IV, and level V grassland coverage. The labeled data set is divided into a training set and a test set, and the data set is randomly divided in a ratio of 8:2, of which 80% of the data is used to train the model and 20% of the data is used to test the performance of the model. Using the above model, the training set is trained, and after the best effect is verified by the test set, the vegetation coverage of the entire county is estimated.

[0064] Figure 3 A schematic diagram of a flow chart of extracting image features by a Block module of a deep learning model provided in an embodiment of the present application, such as Figure 3 As shown in the figure, the remote sensing image of the area to be measured is input into the trained model for semantic segmentation, the Stem module performs preliminary feature extraction, and after generating a shallow feature map, the shallow feature map is input into the Block module. Specifically, it includes two paths: the upper layer is responsible for performing convolution operations and nonlinear transformations, and the lower layer normalizes the data through the Layer Normalization layer. This design not only retains the original feature information, but also enhances the learning ability of the model.

[0065] Specifically: First, the upper layer performs Layer Normalization on the input feature X to obtain the standardized feature X LN ;

[0066] X LN =LayerNorm(X)

[0067] Subsequently, the upper path is divided into two parallel small paths. The upper path 1 performs linear transformation and SiLU activation function processing, and the upper path 2 performs linear transformation, combined with convolution operation and SiLU activation function processing to further extract features. The outputs of the two small paths are fused by element-by-element multiplication and then linearly transformed again to obtain the feature Z of the upper path. upper :

[0068] Z upper =Linear(SiLU(Linear(X LN )).SiLU(Conv(Linear(X LN )))).

[0069] The lower path performs linear transformation, combines the depth-separable convolution DWConv operation and SiLU activation function processing to further extract features and obtain Z lower :

[0070] Z lower =Linear(SiLU(DWConv(Linear(X LN ))))

[0071] Next, the upper Z upper , Lower Z lower Each is added to the original input X through a residual connection to ensure that no information is lost, forming Y upper With Y lower :

[0072] Y upper =Z upper +X

[0073] Y lower =Z lower +X

[0074] Finally, the result Y of the upper and lower layers upper With Y lower By adding and fusing each element, the final output Y of the Block module is generated:

[0075] Y=Y uper +Y lower

[0076] The above summarizes the main calculation process of the Block module from input to output, including key steps such as input normalization, multi-path feature extraction, feature fusion, and residual connection. This structural design helps to enhance the model's ability to understand the multi-level semantic information of remote sensing images while maintaining the simplicity and efficiency of the network.

[0077] S103. Perform statistical analysis on the classification results and calculate vegetation coverage.

[0078] Specifically, the proportion of each type of vegetation in the remote sensing image is calculated, and the overall vegetation coverage in the remote sensing image is calculated. Figure 4 A comparison diagram of the vegetation coverage estimation method based on remote sensing semantic segmentation provided in an embodiment of the present application and the visualization results of the prior art, such as Figure 4 As shown in the figure, "Our" is the result graph output by the method of the present invention, and the rest are the result graphs output by the method of the prior art. It can be seen that the segmentation accuracy of the method of the present invention is better. The comparison results are shown in Table 1:

[0079] Table 1 Comparison of qualitative results of the proposed method and three existing semantic segmentation algorithms

[0080]

[0081] Mean Intersection over Union (MIoU) and Mean Accuracy (MAcc) are the most important indicators for evaluating the performance of semantic segmentation models. The larger the value, the better. As shown in Table 1, compared with three common remote sensing semantic segmentation methods (SegFormer, DCSwin, LSKNet), the method proposed in the present invention has achieved the highest performance in both indicators.

[0082] Figure 5 A comparison diagram of the vegetation coverage estimation method based on remote sensing semantic segmentation and the pixel dichotomy visualization result provided in an embodiment of the present application, such as Figure 5 As shown, a qualitative visualization result comparison is shown. The vegetation estimation method proposed in this application exceeds the value calculated by the current industry standard pixel dichotomy method in terms of accuracy. Table 2 provides a comparison of the accuracy of the existing industry standard vegetation coverage estimation method pixel dichotomy and the method of the present invention in small-scale sample plots. The closer to the actual measurement value, the better. It can be seen that the effectiveness of the method of the present invention.

[0083] Table 2 Comparison of the pixel dichotomy method and the proposed method in terms of accuracy of small-scale plots

[0084]

[0085] Table 3 provides a comparison of the accuracy of large-scale sample plots between the existing pixel dichotomy method and the method of the present invention. The closer to the data value of the third national land survey, the better. Compared with the pixel dichotomy method, the method proposed in the present invention is superior to the pixel dichotomy method in terms of index accuracy and is closer to the actual measurement value. It can be seen that the method of the present invention can adaptively process different types of vegetation coverage in complex environments, thereby improving the generalization ability and stability of the model.

[0086] Table 3 Comparison of the pixel dichotomy method and the method of the present invention in terms of accuracy on large-scale plots

[0087]

[0088] In the method provided by the embodiment of the present invention, by adopting a lightweight design, the network greatly reduces the computational complexity while ensuring the segmentation accuracy, making it suitable for devices with limited resources and improving the flexibility of practical applications. Secondly, the Block module combines deep separable convolution and residual connection, which not only reduces the number of model parameters and computational cost, but also enhances the ability of feature extraction, and improves the accuracy and detail retention of segmentation results. Through the introduction of multi-path feature fusion and SiLU activation function, the network shows good nonlinear expression ability when processing multi-scale remote sensing images, and effectively improves the estimation accuracy of complex objects and vegetation coverage. In addition, the application of layer normalization significantly enhances the stability of the model, ensuring that it can converge efficiently under different data distributions. In general, the present invention greatly improves the segmentation accuracy and computational efficiency of remote sensing images, provides solid technical support for efficient and accurate vegetation coverage estimation, and has broad application prospects.

[0089] Figure 6 A schematic diagram of a vegetation coverage estimation device based on remote sensing semantic segmentation is provided in an embodiment of the present application, such as Figure 6 As shown, the device comprises:

[0090] Image acquisition module 11: used to acquire remote sensing images of the area to be measured;

[0091] Image classification module 22: used to input the remote sensing image into a preset deep learning model to obtain vegetation classification results;

[0092] Statistical analysis module 33: used to perform statistical analysis on the classification results and calculate vegetation coverage.

[0093] The above-mentioned device can execute the vegetation coverage estimation method based on remote sensing semantic segmentation in the above-mentioned embodiment, wherein the image acquisition module 11 executes step S101; the image classification module 22 executes step S102; and the statistical analysis module 33 executes step S103.

[0094] Based on the same inventive concept, an electronic device is also provided in an embodiment of the present application, and the method corresponding to the electronic device may be the vegetation coverage estimation method based on remote sensing semantic segmentation in the aforementioned embodiment, and its principle of solving the problem is similar to that of the method. The electronic device provided in an embodiment of the present application includes: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the methods and / or technical solutions of the aforementioned multiple embodiments of the present application.

[0095] The electronic device may be a user device, or a device formed by integrating a user device and a network device through a network, or may be an application running on the above device. The user device includes but is not limited to various terminal devices such as computers, mobile phones, tablet computers, smart watches, and bracelets. The network device includes but is not limited to network hosts, single network servers, multiple network server sets, or cloud computing-based computer sets, which can be used to implement some processing functions when setting an alarm. Here, the cloud is composed of a large number of hosts or network servers based on cloud computing, where cloud computing is a type of distributed computing, a virtual computer composed of a group of loosely coupled computer sets.

[0096] Figure 7 The structure of a device suitable for implementing the method and / or technical solution in the embodiment of the present application is shown, and the device 1200 includes a central processing unit (CPU, Central Processing Unit) 1201, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM, Read Only Memory) 1202 or the program loaded from the storage part 1208 to the random access memory (RAM, Random Access Memory) 1203. In RAM1203, various programs and data required for system operation are also stored. CPU 1201, ROM 1202 and RAM 1203 are connected to each other through bus 1204. Input / output (I / O, Input / Output) interface 1205 is also connected to bus 1204.

[0097] The following components are connected to the I / O interface 1205: an input section 1206 including a keyboard, a mouse, a touch screen, a microphone, an infrared sensor, etc.; an output section 1207 including a cathode ray tube (CRT), a liquid crystal display (LCD), an LED display, an OLED display, etc., and a speaker, etc.; a storage section 1208 including one or more computer-readable media such as a hard disk, an optical disk, a magnetic disk, a semiconductor memory, etc.; and a communication section 1209 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 1209 performs communication processing via a network such as the Internet.

[0098] In particular, the methods and / or embodiments in the embodiments of the present application may be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes program code for executing the method shown in the flowchart. When the computer program is executed by the central processing unit (CPU) 1201, the above functions defined in the method of the present application are executed.

[0099] Another embodiment of the present application further provides a computer-readable storage medium having computer program instructions stored thereon, wherein the computer program instructions can be executed by a processor to implement the methods and / or technical solutions of any one or more embodiments of the present application described above.

[0100] Specifically, the present embodiment may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device, or device.

[0101] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer readable signal media may also be any computer readable medium other than a computer readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0102] The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0103] Computer program code for performing the operation of the present application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., using an Internet service provider to connect through the Internet).

[0104] The flow chart or block diagram in the accompanying drawings shows the possible architecture, function and operation of the equipment, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated system for hardware that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0105] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0106] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or page components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0107] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0108] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to perform some steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), disk or optical disk and other media that can store program codes.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

[0110] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in a device claim can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.

Claims

1. A vegetation coverage estimation method based on remote sensing semantic segmentation, characterized in that: include: Acquire remote sensing images of the area to be measured; Inputting the remote sensing image into a preset deep learning model to obtain a vegetation classification result; The classification results were statistically analyzed and the vegetation coverage was calculated.

2. The vegetation coverage estimation method based on remote sensing semantic segmentation according to claim 1 is characterized in that: Inputting the remote sensing image into a preset deep learning model to obtain a vegetation classification result further includes: Build a convolutional neural network model based on the encoder-decoder architecture; Inputting the remote sensing image into the convolutional neural network model for semantic segmentation; According to the semantic segmentation result, the vegetation type in the remote sensing image is output.

3. The vegetation coverage estimation method based on remote sensing semantic segmentation according to claim 2 is characterized in that: The step of inputting the remote sensing image into the convolutional neural network model for semantic segmentation further comprises: Input the remote sensing image into the Stem module for preliminary feature extraction to generate a shallow feature map; The shallow feature map is subjected to convolution calculation through multiple Block modules to extract image features; The processed feature map is restored to its original resolution and the semantic segmentation result is generated.

4. The vegetation coverage estimation method based on remote sensing semantic segmentation according to claim 3 is characterized in that: The step of performing convolution on the shallow feature map through a plurality of Block modules to calculate image features further includes: Normalize the input shallow feature map; The activation function SiLU is used to perform convolution calculation on the normalized features to extract image features.

5. The vegetation coverage estimation method based on remote sensing semantic segmentation according to claim 4 is characterized in that: The step of using the activation function SiLU to perform convolution calculation on the normalized features to extract image features further includes: In the upper path, the normalized features are linearly transformed and convolved, and the SiLU activation function is used to obtain the upper path features; In the lower path, the normalized features are linearly transformed and the depth-wise separable convolution DWConv is calculated, and the SiLU activation function is used to obtain the lower path features; The upper layer path features and the lower layer path features are respectively residually connected and added to the original input features to obtain complete upper layer path features and lower layer path features. The complete upper-layer path features and lower-layer path features are added and fused element by element to obtain the image features.

6. The vegetation coverage estimation method based on remote sensing semantic segmentation according to claim 1 is characterized in that: The statistical analysis of the classification results to calculate the vegetation coverage further includes: The proportion of each type of vegetation in the remote sensing image is calculated, and the overall vegetation coverage in the remote sensing image is calculated.

7. The vegetation coverage estimation method based on remote sensing semantic segmentation according to claim 1 is characterized in that: The method further comprises: The training samples of the deep learning model are updated regularly, and the model parameters are updated using the updated training samples.

8. A vegetation coverage estimation device based on remote sensing semantic segmentation, characterized in that: include: Image acquisition module: used to obtain remote sensing images of the area to be measured; Image classification module: used to input the remote sensing image into a preset deep learning model to obtain vegetation classification results; Statistical analysis module: used to perform statistical analysis on classification results and calculate vegetation coverage.

9. A vegetation coverage estimation device based on remote sensing semantic segmentation, the device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

10. A computer readable medium having computer program instructions stored thereon, wherein the computer program instructions can be executed by a processor to implement the method according to any one of claims 1 to 7.

Images